Does debt slow growth?

Good arguments on both sides in this debate

The econoblogosphere and many newspapers erupted last week over errors in “Growth in a Time of Debt,” a politically important research paper by Carmen Reinhart and Ken Rogoff.

Reinhart and Rogoff looked at the experience of many countries, over long swaths of time, and concluded that countries with low levels of public debt had much stronger growth performance than those with high debt. They observed that the big hit on growth was most striking for nations that had public debt above 90% of gross domestic product (GDP), as opposed to those with debt-to-GDP ratios below 60%.

Reinhart-Rogoff has often been cited in the “austerity” debates in the U.S. Congress. There, many Republicans, who claim to want lower federal deficits and a lower government debt-to-GDP ratio, are pitted against most Democrats, who fear that austerity would undermine their jobs and growth agenda. Similarly, “austerity” has become a political battle cry in countless other jurisdictions struggling with the aftermath of financial and economic crises.

As we now know, the study’s conclusions depend on spreadsheets that include data and coding errors. Corrected for these, the results are much weaker. More fundamentally, Reinhart-Rogoff ’s techniques do not address well the question of “reverse causality.” Does high debt cause low growth, or does low growth, meaning lower-than-otherwise GDP, subsequently cause the debt-to-GDP ratio to be higher-than-otherwise?

Economists have reason to doubt whether cross-country analyses are terribly useful in answering questions like these. Paraphrasing the Twitterterse words of Kevin Milligan of the University of British Columbia: plus or minus one Cape Verde in your dataset, and all your conclusions change. And even if the data and the analyses are right, does that mean the conclusions apply here, now, in this country or that, and in its circumstances?

The assumption of those who take such studies at face value, and of their current antagonists, is that the answer to the last question is yes, the conclusions apply here and now. And both sides accept that the results rest on a reliable theoretical foundation.

So … let’s play that game too, and ignore existential doubts. Assume — and if you believe in these kinds of analyses, this implies a leap of logic, and if you do not, a leap of faith — that the Reinhart-Rogoff approach could be useful, if corrected for data and coding errors and the possibility of reverse causality.

Arindrajit Dube, of the University of Massachusetts, Amherst, did the math that lets us do that. He finds that for countries with below about 75% of GDP, less debt means more

The growth we could be forgoing amounts to Canada’s total GDP today

growth. For very high debt countries, the debt’s negative impact on growth is not nearly so evident (see figure, for which I thank Professor Dube).

What if this relationship applied to Canada, now? Canada’s federal plus provincial debt-to-GDP ratio is about 65%; what if it were 25% from here on? Following Reinhart-Rogoff and Professor Dube’s approach, growth could be higher in Canada by about 0.4% annually.

That doesn’t sound like a lot. So attach some dollars to it, let them grow over time, and account for the fact that a dollar now is worth more to us than a dollar later.

Under a pretty tame set of growth and discounting assumptions, that 0.4% in forgone growth becomes gigantic, in dollar and relative terms, in just a few years. The growth we could be forgoing, at 0.4% in real annual terms, should it persist over 23 years, amounts to Canada’s total GDP today. Or, paying down debt, today, to the tune of 40% of Canada’s GDP, would buy a stream of 23 years of income growth worth more than 100% of GDP now, today.

Does that mean we should pull out all the stops to shrink Canada’s federal and provincial debt, to 25% of GDP? Probably not.

The critics of quick deficit reduction have good arguments. Trivially, national income accounting says that deficit-financed government spending, this year, makes GDP bigger than otherwise, this year. It also keeps people employed, improving skills and long-run job continuity. And it puts money in people’s hands, which is good — if not necessarily a source of long-run growth. More importantly, winning contracts now builds business cash flow, and confidence; business will invest in more projects, create more jobs, and put the country on a better growth trajectory. That’s what bigger deficits buy.

But hold the phone. Is it safe to assume government infrastructure spending on, say, ice rinks builds people’s skills, and that past job continuity matters when trades-folk need work next year and the year after and after that? And is it safe to assume that big government deficits make business confidence go up, not down, and that deficit spending does not just buy holes in the ground or in the long-run balance sheet? These don’t seem sure bets.

So what do we know? First, the battle remains among those who want to continue to run monstrous fiscal deficits, and those who pursue only gigantic deficits, and a smaller contingent who would prefer smaller deficits or none.

In this battle, we tend to forget the bigger growth question. Do governments spend wisely, on things that are likely to be productive for the long haul?

What else we know is not much, but it is important. Governments that have lower debt than others will be better placed to face the next shock, to cope with an aging population, and to otherwise deal with whatever lays in front of us. They might even deliver better growth outlooks.